From 138d540b9adc3f141851a76cbdcda176d8567496 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E2=80=9CDaniel=20Griffiths=E2=80=9D?=
<Danielgriffiths1790@gmail.com>
Date: Mon, 8 Jan 2024 15:34:30 +0000
Subject: [PATCH] feat: added more tfidf tests
---
tests/unit/encoders/test_tfidf.py | 25 ++++++++++++++++++++++++-
1 file changed, 24 insertions(+), 1 deletion(-)
diff --git a/tests/unit/encoders/test_tfidf.py b/tests/unit/encoders/test_tfidf.py
index 5aa8dfc8..21524c91 100644
--- a/tests/unit/encoders/test_tfidf.py
+++ b/tests/unit/encoders/test_tfidf.py
@@ -39,7 +39,7 @@ class TestTfidfEncoder:
isinstance(sublist, list) for sublist in result
), "Each item in result should be a list"
- def test_call_method_no_docs(self, tfidf_encoder):
+ def test_call_method_no_docs_tfidf(self, tfidf_encoder):
with pytest.raises(ValueError):
tfidf_encoder([])
@@ -60,3 +60,26 @@ class TestTfidfEncoder:
def test_call_method_with_uninitialized_model(self, tfidf_encoder):
with pytest.raises(ValueError):
tfidf_encoder(["test"])
+
+ def test_call_method_no_docs(self, tfidf_encoder):
+ with pytest.raises(ValueError, match="No documents to encode."):
+ tfidf_encoder([])
+
+ def test_compute_tf_no_word_index(self, tfidf_encoder):
+ with pytest.raises(ValueError, match="Word index is not initialized."):
+ tfidf_encoder._compute_tf(["some docs"])
+
+ def test_compute_tf_with_word_in_word_index(self, tfidf_encoder):
+ routes = [
+ Route(
+ name="test_route",
+ utterances=["some docs", "and more docs", "and even more docs"],
+ )
+ ]
+ tfidf_encoder.fit(routes)
+ tf = tfidf_encoder._compute_tf(["some docs"])
+ assert tf.shape == (1, len(tfidf_encoder.word_index))
+
+ def test_compute_idf_no_word_index(self, tfidf_encoder):
+ with pytest.raises(ValueError, match="Word index is not initialized."):
+ tfidf_encoder._compute_idf(["some docs"])
--
GitLab